Historical backtesting
Test a strategy against historical market data with configurable capital, costs, slippage, frequency, and risk assumptions.
Research first. Execution stays with you.
Develop trading ideas with evidence, not guesswork.
Design strategies with AI, test them on historical data, inspect the assumptions, and export the code—all in a workspace built for careful iteration.
Strategy version 04
Momentum with regime filter
Annual return
18.4%
Max drawdown
-9.7%
Sharpe ratio
1.42
Point-in-time snapshot locked
Commission and slippage included
Out-of-sample window reserved
Illustrative figures from a sample run, not a performance claim.
Connecting to market data
Reference prices only — AlgoTrader AI does not execute trades.
Watch it work
One idea, three iterations, every number recomputed
A recorded research session: a breakout thesis becomes rules, the rules get tested, and each revision is re-run against the same sample. The metrics beside the chart are computed from the trades drawn on it.
BTC / USD
Daily candles · 240-session sample · simulated data
v1v2v3
Awaiting rulesv1 Buy 8-bar breakouts, exit on a 4-bar low.
Return
—
Sharpe
—
Max drawdown
—
Win rate
—
Trades
—
8-bar highEntryExit
Strategy copilot
Breakout research · 240 daily sessions
Breakouts on BTC look obvious in hindsight, but every time I trade one I get chopped up. Can we find out whether the signal is actually worth anything?
Ask for another revision…
A disciplined research loop
From thesis to testable strategy
Keep the idea, implementation, dataset, and result connected as your strategy evolves.
Iterative AI development
Turn a plain-language thesis into concrete rules, then refine the strategy and compare each version without losing the research trail.
Temporal safeguards
Point-in-time data snapshots, explicit date ranges, and recorded execution assumptions help you spot look-ahead bias and leakage.
Portable code export
Export the strategy code and documented assumptions for review, extension, or use in your own research environment.
Run your algorithm
Tested code, ready to run
Every strategy you backtest and iterate on is real code that has already been run. Take it with you and run it wherever you like.
Download it. Run it anywhere.
Each strategy version you backtest is backed by real code. Export it as a self-contained package with its configuration and expected results, then run it on your own machine or server and confirm it reproduces the in-app run.
- Every exported version is code that already ran through a backtest
- Python 3.11 standard library only, no dependencies to install
- Runs on your laptop, a VPS, or any server you control
momentum-regime-v04.tar.gz
Tested- strategy.pyThe exact version you backtested
- strategy_runtime/Strategy contract 1.0
- data_provider.pyPlug in your own market data
- configuration.jsonFrozen capital, costs, and dates
- expected-result.jsonTrades and metrics from the app run
- reproduce.pyChecks your run matches, trade for trade
- manifest.jsonVersions and source SHA-256
your-server
$ tar -xzf momentum-regime-v04.tar.gz
$ python3 cli.py --verify
$ python3 reproduce.py
Reproducibility check passed
Free to read
The Trading Library
153 in-depth guides covering every major trading strategy, indicator, chart pattern, risk framework, and backtesting method. Written so someone starting from zero can follow along.
Start here
What is a trading strategy?
Rules for entry, exit, and size, explained from zero, with a full worked example.
Strategy types
Every major strategy family compared
Trend, mean reversion, momentum, breakout, arbitrage, carry, and event driven.
By market
Best strategies by asset class
What actually fits stocks, options, futures, forex, crypto, and commodities.
Risk
Position sizing and risk management
The arithmetic that decides whether a strategy survives its own drawdowns.
Validation
Backtesting without fooling yourself
Survivorship bias, look-ahead bias, overfitting, and realistic cost modelling.
Indicators
Technical indicators explained
What each one actually measures, and how to turn it into a testable rule.
No live trading or order execution
AlgoTrader AI is a research and backtesting workspace. It does not connect to a brokerage, place trades, or manage money. Simulated results are not a promise of future performance.